Overview

We’re excited to share some major updates to Genderfluent. This release includes adding voice formant analysis, improved pitch and gender processing, and some subscription updates. Here’s a detailed look at what’s new!

Vocal Formants

Formants are the resonant frequencies produced by the shape of your vocal tract, such as vocal track length, tongue position, jaw position, etc. These frequencies are key to the unique timbre of your voice and play a crucial role in how others perceive your voice, particularly in relation to gender. Analyzing formants can provide more actionable information for voice training and help you understand how to adjust your voice to achieve your desired sound.

Learn more about formants in our blog post here.

Below is a screenshot of an example formant analysis. The top graph is pitch and bottom graph shows the formants, which has three measurements called F1, F2, and F3. During the recording, pitch was kept constant while altering the vocal tract shape to change the formants.

Important note about privacy:

Given the computational complexity of formant analysis, this algorithm is too resource-intensive to be performed on most user devices. Instead, when you record your voice, the audio is securely sent to our backend servers. There, it is processed and the results are sent back to your device.

In the future, we do hope to move formant processing to on-device to enhance user privacy. While there currently are some formant analysis algorithms that are fast enough to run on device, their accuracy is lower. We want to provide the best user experience possible, so we choose to offer the state-of-the-art formant analysis.

More support for gender estimation

On a related note, we understand that not all devices are capable of performing gender estimation due to device or software limitations. To address this, we’ve introduced an optional feature where users can now choose to stream their audio to our backend for gender estimation. So if your device is unable to handle the processing requirements, you should see a notification prompting you to enable this feature.

Pitch improvements

In this update, we’ve also enhanced our pitch detection algorithm. You’ll notice more precise pitch tracking, especially in challenging scenarios such as rapid pitch changes or background noise. We’ll continue to refine the algorithm to ensure accurate and reliable pitch analysis.

Policy changes

Audio Retention Policy

We’ve updated our audio retention policy to better align with our users’ needs. For users who have subscribed to our service at any point, we will now retain their audio data that exceeds what’s allowed under the free plan for up to 5 years after their subscription ends. This allows for continuity in service should they decide to resubscribe and ensures their previous data is available for reference or comparison.

Temporary Discount

As a thank you to our users, we’re offering a temporary 30% discount on subscriptions. Just use the code VOICE2024, which is valid until September 20th. This is a great opportunity to take advantage of all the new features and enhancements we’ve introduced.